Indian Refrigerator Industry

...

[pic 7]

Whole sale price index

This checks the change in prices of market basket of goods and services at a whole sale stage. This statistical value is calculated by taking a sample of representative items, which lie in a fixed market basket and their prices are collected periodically to measure the change. This is primary measure for checking the inflation.

[pic 8]

Sensex

Sensex indicates the relative prices of shares on the Bombay Stock Exchange. Sensex is the BSE 30 share index. Changes in Sensex can influence the sale and demand of refrigerators. In our analysis, we have used the mean of the opening and closing values for each quarter from 1998 to 2010.

[pic 9][pic 10]

Methodology

In this analysis, three kinds of models have been used for the forecasting of the sale of refrigerators in India:

- Regression Models

- Exponential Smoothing Models

- Decomposition Models

Both simple linear and multiple linear regression models have been used. In exponential smoothing models, simple exponential smoothing, EWS Holt model and EWS Holt Winters model have been used. In decomposition models, additive and multiple decomposition models have been used.

Regression Models

Regression models involve the forecasting of dependent variables based on historical data of dependent and independent variables. However, regression models do not imply causality but denote correlation.

To forecast sale of refrigerators, both simple and multiple linear regression have been used.

Simple linear regression

The dependent variable here is sales of refrigerators and the independent variables are IIP, Real GDP, short-term interest rates, WPI, Real Private Consumption and Sensex. All figures have been measured quarterly from 1998-2010. Variables that exhibited a correlation above .75 have been chosen

Simple linear regression can be represented as:

[pic 11]

Here y represents the dependent variable, x is the independent variables and e is the error term while a is a constant.

Y variable

X variable

r

Intercept

Slope

R^2

Adj. R^2

DW

MAPE

Total Refrigerator Sales

IIP

0.877021

-974.393

21.0226

0.769166

0.76455

1.971184

0.224019887

Total Refrigerator Sales

Real GDP

0.884539

-1198.5

0.067283

0.78241

0.778058

2.015732

0.222241866

Total Refrigerator Sales

Interest Rate (Short term)

0.204874

2729.4

-134.709

0.041973

0.022813

-1.17795

0.703695847

Total Refrigerator Sales

WPI

0.868728

-2020.28

32.06205

0.754688

0.749782

1.885706

0.254583452

Total Refrigerator Sales

Real Private Consumption

0.884133

-1404.47

5.523236

0.781691

0.777325

2.044932

0.240220022

Total Refrigerator Sales

SENSEX

0.775144

247.8863

0.112216

0.600848

0.592864

1.495214

0.3264459

Multiple Linear Regression

In multiple linear regression, we have used a combination of independent variables to arrive at an equation predicting the values of the dependent variable.

The multiple linear regression model can be represented as: Yi = a + 1 X1i + 2 X2i + 3 X3i + … + k Xki + i [pic 12]

Here y is the independent variable, x1,2,3 are dependent variables and e is the error term.

X Variable

Adjusted R^2

DW

MAPE

IIP,